Episode 38
Can the Human Line Project, a Nonprofit, Solve AI Delusion, Psychosis?
By David Stephen who looks at AI Delusion in this article. There is a recent [March 27, 2026] spotlight on Sky News, The AI delusion problem is bigger than we thought, stating that, "A new study from Anthropic and the University of Toronto analysed 1.5...
Irish Tech News Audio Articles · Simon Cocking
April 2, 20266m 50s
Audio is streamed directly from the publisher (cdn.beyondwords.io) as published in their RSS feed. Play Podcasts does not host this file. Rights-holders can request removal through the copyright & takedown page.
Show Notes
By David Stephen who looks at AI Delusion in this article.
There is a recent [March 27, 2026] spotlight on Sky News, The AI delusion problem is bigger than we thought, stating that, "A new study from Anthropic and the University of Toronto analysed 1.5 million conversations with the AI chatbot Claude, revealing rare but concerning cases of what some are calling "AI psychosis" and what researchers describe as "reality distortion". What happens when AI starts pulling people away from reality and even encourages them to act on distorted beliefs?"
The challenges of AI Delusion
There is recent [March 26, 2026] analysis on The Guardian, Marriage over, €100,000 down the drain: the AI users whose lives were wrecked by delusion, stating that, "Last year, the first support group for people whose lives have been derailed by AI psychosis was formed. The Human Line Project has collected stories from 22 countries. They include 15 suicides, 90 hospitalisations, six arrests and more than $1m (£750,000) spent on delusional projects. More than 60% of its members had no history of mental illness."
"Modern AI chatbots built on large language models – advanced AI systems – are trained on enormous datasets to predict word sequences: it's a sophisticated system of pattern matching. Yet even knowing this, when something non-human uses human language to communicate with us, our deeply ingrained response is to view it – and to feel it – as human. This cognitive dissonance may be harder for some people to carry than others."
The Human Line Project
Having a support group against AI delusion and psychosis is admirable, given that AI companies are doing nothing exponential against both, seeing how the problem persists, in spite of adjustments they have announced over the last few months.
All AI chatbots have disclaimers, ensuring the user is aware that AI is not an organism. But with AI, sycophantic, then a penetrative ability to use human language and express understanding across human knowledge areas, it appeals to several minds, and for some, excessively. And this is good for business.
Saying AI should not be sycophantic is like saying that social media should not have doom scrolling. That AI chatbots are like digital concierge, or servile, with how excellent they are, makes people loiter. It is a vital part of the value offering.
This means that it is unlikely that there would ever be some encompassing software engineering solution to AI psychosis. So, expecting thorough regulation, or some reduction in sycophancy or some algorithmic tweak may not be deep enough to make much difference against AI psychosis.
Now, the option is to go to the source, the human mind, to seek out what it might mean that AI is causing or reinforcing delusions. Simply, how will it be possible to have a conceptual design of the mind, with relays and destinations to display what the AI might be doing?
This conceptual display [or theoretical neuroimaging] can be a separate application where keywords from a chat can be pasted, so that it is possible to simulate relays and destinations the mind went, then show those skipped [caution, consequences], as well as the types of relay interchanges, which means using reality paths for non-reality like imagination.
The model is based on the postulation in Conceptual Biomarkers and Theoretical Biological Factors for Psychiatric and Intelligence Nosology.
For Profit
It is unlikely that the solution to AI delusion and psychosis would be effective or competitive as a nonprofit. The commercial interests are in trillions of dollars, by market cap. There is no way to match them, with donations or partnerships or a few reports in the news.
The mind safety display against AI psychosis would be for profit, having subscribers [be able to] access features that would be useful to stay in check.
There could be a free version, only for general sample relays. As profits are pursued aggressively and soars, it will possible to develop better iterations an...
There is a recent [March 27, 2026] spotlight on Sky News, The AI delusion problem is bigger than we thought, stating that, "A new study from Anthropic and the University of Toronto analysed 1.5 million conversations with the AI chatbot Claude, revealing rare but concerning cases of what some are calling "AI psychosis" and what researchers describe as "reality distortion". What happens when AI starts pulling people away from reality and even encourages them to act on distorted beliefs?"
The challenges of AI Delusion
There is recent [March 26, 2026] analysis on The Guardian, Marriage over, €100,000 down the drain: the AI users whose lives were wrecked by delusion, stating that, "Last year, the first support group for people whose lives have been derailed by AI psychosis was formed. The Human Line Project has collected stories from 22 countries. They include 15 suicides, 90 hospitalisations, six arrests and more than $1m (£750,000) spent on delusional projects. More than 60% of its members had no history of mental illness."
"Modern AI chatbots built on large language models – advanced AI systems – are trained on enormous datasets to predict word sequences: it's a sophisticated system of pattern matching. Yet even knowing this, when something non-human uses human language to communicate with us, our deeply ingrained response is to view it – and to feel it – as human. This cognitive dissonance may be harder for some people to carry than others."
The Human Line Project
Having a support group against AI delusion and psychosis is admirable, given that AI companies are doing nothing exponential against both, seeing how the problem persists, in spite of adjustments they have announced over the last few months.
All AI chatbots have disclaimers, ensuring the user is aware that AI is not an organism. But with AI, sycophantic, then a penetrative ability to use human language and express understanding across human knowledge areas, it appeals to several minds, and for some, excessively. And this is good for business.
Saying AI should not be sycophantic is like saying that social media should not have doom scrolling. That AI chatbots are like digital concierge, or servile, with how excellent they are, makes people loiter. It is a vital part of the value offering.
This means that it is unlikely that there would ever be some encompassing software engineering solution to AI psychosis. So, expecting thorough regulation, or some reduction in sycophancy or some algorithmic tweak may not be deep enough to make much difference against AI psychosis.
Now, the option is to go to the source, the human mind, to seek out what it might mean that AI is causing or reinforcing delusions. Simply, how will it be possible to have a conceptual design of the mind, with relays and destinations to display what the AI might be doing?
This conceptual display [or theoretical neuroimaging] can be a separate application where keywords from a chat can be pasted, so that it is possible to simulate relays and destinations the mind went, then show those skipped [caution, consequences], as well as the types of relay interchanges, which means using reality paths for non-reality like imagination.
The model is based on the postulation in Conceptual Biomarkers and Theoretical Biological Factors for Psychiatric and Intelligence Nosology.
For Profit
It is unlikely that the solution to AI delusion and psychosis would be effective or competitive as a nonprofit. The commercial interests are in trillions of dollars, by market cap. There is no way to match them, with donations or partnerships or a few reports in the news.
The mind safety display against AI psychosis would be for profit, having subscribers [be able to] access features that would be useful to stay in check.
There could be a free version, only for general sample relays. As profits are pursued aggressively and soars, it will possible to develop better iterations an...